Image List Split Head Tail

Some video workflows are built from "transition" clips: you have a sequence of keyframe images - a door closing, a camera pan, a scene change - and you need a short video that morphs from image 1 to image 2, then from 2 to 3, and so on. ImageListSplitHeadTail exists to turn that image list into ready-made start frame / end frame pairs. Give it 5 images and it produces 4 pairs - (1→2), (2→3), (3→4), (4→5) - each one ready to feed an image-to-video model as its conditioning.

How it works

The math is deliberately simple: N input images give N-1 pairs, where each pair is (frame[i], frame[i+1]). You then select a range of those pairs with two knobs:

• start_index - which pair to begin at (0 = the first pair, per the author's tooltip). Start mid-list if you only want the tail transitions.
• index_count - how many pairs to take. Default 100, which means "everything" for any realistic list.

So with input [1,2,3,4]: start_index=0, index_count=1 gives you just (1,2); start_index=1, index_count=2 gives (2,3),(3,4). The tooltips spell this out, and it's the one behavior worth internalizing before you wire it up.

Outputs

• start_frames (IMAGE list) - the "from" images.
• end_frames (IMAGE list) - the matching "to" images, same length.
• final_end_frame (single IMAGE) - the last frame of the whole set, so you can keep the timeline continuous even if you only processed some pairs.

The pattern to use it: run each (start_frames[i], end_frames[i]) pair through an image-to-video node as first/last frame conditioning, then splice the rendered clips in order. Because the pairs share boundary frames, the transitions link up seamlessly - the last frame of clip 1 is the first frame of clip 2.

Installing it

Part of ComfyUI-ListHelper:

cd ComfyUI/custom_nodes
git clone https://github.com/dseditor/ComfyUI-ListHelper

Restart ComfyUI; find it under ListHelper/Tools. No models, no dependencies beyond the pack's base requirements. ComfyUI Manager: search "ComfyUI-ListHelper".

The honest caveats

Two things to know before you build a whole pipeline on it. First, it needs at least 2 images or it raises a clear error - a single frame can't make a transition. Second, this node only pairs frames; it doesn't interpolate or generate the in-between motion. The actual video is produced by whatever image-to-video node you feed the pairs into, so your results are only as good as that model's transition quality. What this node gives you is the plumbing - correct, matching head/tail frames every time, no off-by-one bugs in a workflow you'd otherwise hand-assemble. For the transition-driven video pipelines this pack is clearly built around, that plumbing is the difference between a workflow that works and one that silently drops pairs.